Predictive processing, predictive coding - a BPS theory of causation

[dave30th]

Moderation note:
we also have a Members Only thread discussing this topic - here
posts have been moved from the Paul Garner thread

***************************

For some reason some psychological medicine people have misunderstood a basic neuroscience theory and got it back to front.

Jo, can you explain what you mean here? What is back to front exactly?

 

[Jonathan Edwards]

Jo, can you explain what you mean here? What is back to front exactly?

Predictive coding says that you experience a patch of cloth as red because your retina is getting cyan input all around and expecting the patch to be cyan but it is slightly less cyan - so it is read as the opposite, which is red. Pat Merton got someone to grab his eyeball with forceps and move it to the left. He sensed the world turning to the right, as in vertigo. If he had moved his eyes himself his brain would have predicted the image change exactly as it did and experienced C-C = 0 = 'no change'. With the forceps 0 - C = -C.

In other words some brain cells subtract the expectation from the immediate input and present you with the difference as 'what is happening'. This might seem a grand new theory but even in the seventeenth century there were people who understood that this is the only way things could possibly work anyway. All our experiences of the world have to be based on differences. The deeper reasons behind that we can leave for now. (It is actually quite obvious if you think about it for a bit.)

So if you are not expecting input from pain fibres and you get input from pain fibres you get

PAIN

If you are expecting input from pain fibres and you don't get input you get

RELIEF

The old unhelpful beliefs theory was that people with ME/CFS feel bad with activity because they are expecting to. But try that in predictve theory:

Expecting to feel bad. No bad signals coming in:

RELIEF, BETTER THAN EVER

The bizarre thing is that the theory is stil the old one - that symptoms occur because they are expected. In the old simple theory that was vaguely plausible. In predictive coding it is the opposite of what the theory predicts. And almost nobody selling the predictive coding theory seems to have noticed, not even Mark Edwards who seems to have started it.

We did have one neurologist here from the antipodes who agree that what is happening must actually be the opposite of a predictive coding mechanism. I even think Levinovitz agreed that it didn't work (or somebody else recently). People who know about Pavlov (who is much more relevant here) see that it doesn't work. But the barrel organ plays and the monkey dances and people pay their money.

 

[Axel]

I can confirm that. My symptom-free periods are pure Nirvana.

 

[Sean]

I think the substantial degree of unpredictability, and disproportionality, of the symptom response to behavioural manipulations, especially early on in the patient's experience of it, almost on its own falsifies the predictive hypothesis.

I can have very good or very bad days, without changing my basic behaviour pattern at all. Where is the critical predictive component, the causal link?

Citing 'subconscious' processes, as the FND crowd so casually love to do, is not an explanation in any sense. It is a straight dodge and fudge. It just moves the problem to somewhere even more unexplained, and unexplainable, and not amenable to conscious manipulation.